I analyze the impact of changes to collateral value on borrowers' default decision on auto loans using two types of natural experiments in Sri Lanka. Changes in vehicle import tax rates and loan-to-value ratio caps on auto loans generated plausibly exogenous variation in the resale value of vehicles already pledged as collateral. Using proprietary auto loan performance data, I estimate that a 10% drop in the collateral value corresponds to a 44% increase in the default rate. I also find that collateral value is more important for borrowers with higher outstanding loan balances.
Interest Rate, Consumption, and Default: Evidence from Small Changes in the National Mortgage Rate (with Don Carmichael and Kevin Roshak)Boulder Summer Conference 2018 (Poster Session) | FMA 2017 (Semifinalist for best paper) | IBEFA Summer Meeting 2017 (co-author)
Modest differences in the interest rate at loan origination can have long-lasting effects on mortgagors. We use monthly fluctuations in the national mortgage rate at loan origination to study small changes in interest rates across home purchases made in the same year, in the same area, and which eventually reach similar levels of negative equity. A 50bp higher national rate at origination corresponds to an extra $550 in payments per year and, during the bust, an increase in defaults of 68-88 bp within 12 months of reaching negative equity. The effect is large relative average default rates of 3.78% (5.39%) for homes with 10% (30%) negative equity. Consistent with liquidity constraints, the magnitude of the effect is relatively constant across different levels of negative equity. The national mortgage rate is not correlated with worse borrower credit quality. During the boom, smaller mortgage payments result in increased consumption of non-durables and services from 2001-2007, while total expenditure is unchanged. If intermediaries resist large concessions to borrowers, small concessions may be more effective.
Sunk-Cost Fallacy and Seller Behavior in the Housing Market (with Vijay Yerramilli)ARES 2018 | FMA 2018 - Scheduled | MFA 2018
We use a unique feature of California's property tax system to empirically identify the effect of selling homeowners' past property tax payments on their choice of listing price. Although past property taxes are sunk costs, we find that they have a significant positive effect on the sellers' choice of listing price, which is inconsistent with rational models of decision making. This effect is stronger when sellers expect to sell at a loss relative to their purchase price, for high-valued properties, for zip codes with more active housing markets, and for sellers with less mortgage debt. Overall, our results suggest that the sunk-cost fallacy affects seller behavior in the housing market.
In this paper I examine how servicer characteristics affect foreclosure rates of delinquent securitized mortgages. Using a sample of mortgages originated by IndyMac Bank and Countrywide Home Loan Servicing that underwent a change in servicer during the financial crisis period, I first show using a difference-in-difference methodology that the servicer has a significant effect on foreclosure rates. Consistent with the idea that information asymmetry affect foreclosure rates, I find a significantly lower foreclosure rate among mortgages where the originator also acts as servicer. Finally, foreclosure rates are ceteris paribus higher for servicers that experienced high loan growth in the past, which suggests that servicer capacity constraints affected foreclosure rates.
Work in Progress
Risk of Flooding and Climate Gentrification: Evidence from Hurricane KatrinaThis paper investigates whether low-income households and minorities are forced into areas more vulnerable to flooding due to sea level rise, the process known as inverse climate gentrification. The setting for this study is the affected counties in Mississippi before and after being hit by Hurricane Katrina in 2005. I use the mortgage applications in census tracts that were impacted by Hurricane Katrina as the treatment sample and controls sample consists of mortgage applications in unaffected census tracts. Preliminary results suggest that family income reported in mortgage applications for owner-occupied new home purchases in affected areas was 5.8% lower compared to unaffected areas and the fraction of non-white applicants increased by 5% in affected areas.